CFAR DETECTION OF FLUCTUATING TARGETS IN SPATIALLY CORRELATED K-DISTRIBUTED CLUTTER

The problem of single pulse CFAR detection of Rayleigh fluctuating targets in K-distributed sea clutter is addressed. The performance of three types of well known CFAR processors, namely the cell averaging (CA), cell averaging greatest of (CAGO), and order statistic (OS) CFAR processors, is examined under varying conditions of clutter spatial correlation for several cases of desired false alarm probability and the number of reference cells used in the CFAR processor. Curves for the detectability loss due to the 'spikiness' of the clutter are presented and values for the additional loss due to CFAR thresholding are tabulated. The effects of incorrectly estimating the clutter shape parameter v are investigated. A model is presented to facilitate numerical analysis of detection performance under conditions of partial correlation of the clutter modulation process. Sample results are presented and it is shown that for the CFAR processors analysed here detection performance will not in general be dramatically improved relative to the worst case condition of spatially uncorrelated clutter.